Sheet counting & crack detection
Problem. Manual tallies of stacked roofing sheets are slow and error-prone, and cracked sheets slip through.
What it does. Counts sheets and flags cracks from a camera feed, live on Jetson at the edge.
Cameras and drones capture vast amounts of video, and most of it is never used. We turn it into counts, defects, safety alerts and decisions — in real time, at the edge, even where connectivity is poor.
We design, build, integrate, customise and support the complete system — drones and edge devices, a 4-tier AI platform and sector-specific models — so customers buy outcomes, not components.
Runs those models where the data is — far edge, edge, fog and cloud — and keeps working offline.
PlatformDrones, rovers and edge devices that capture the data, including a crash-tolerant tactical UAV.
Tactical UAVRecordings from our own working prototypes. Every one runs on the same Sahasraksha edge-to-cloud platform.
Problem. Manual tallies of stacked roofing sheets are slow and error-prone, and cracked sheets slip through.
What it does. Counts sheets and flags cracks from a camera feed, live on Jetson at the edge.
Problem. Roll calls eat into class and shift time, and proxy attendance is hard to catch.
What it does. Recognises each person on Jetson and marks them Present in the ERP — contactless.
Problem. Early signs that a student is struggling are easy to miss on a large campus.
What it does. Consent-based wellness score from meal, presence and emotion signals; staff review every flag.
Problem. Safety officers cannot watch every worker on a busy site, every hour.
What it does. Flags missing hard-hats, masks and safety vests in real time, with image evidence.
Industrial sites need answers in milliseconds, on poor links, without shipping every frame upstream. That is why our models run at the edge first.
Edge computing market: $168B (2025) → $249B (2030). Source: MarketsandMarkets, August 2025.
Our baseline drone already flies autonomous waypoint missions. A tensegrity airframe that absorbs crashes has been built and drop-tested, and LiDAR-SLAM navigation is in development — converging into one UAV for border zones and other high-risk terrain.
Installation and on-site commissioning · remote monitoring and model updates · annual maintenance plans · 24/7 support options.
We run a live demonstration on your camera feeds or drone footage, then a 30-day proof of concept at your site — before you commit.